JLang wrote:
] Abstract
] This paper considers the problem of using wireless LAN
] location-sensing for security applications. Recently, Bayesian
] methods have been successfully used to determine location from
] wireless LAN signals, but such methods have the drawback that
] a model must first be built from training data. The
] introduction of model error can drastically reduce the
] robustness of the location estimates and such errors can be
] actively induced by malicious users intent on hiding their
] location. This paper provides a technique for increasing
] robustness in the face of model error and experimentally
] validates this technique by testing against unmodeled
] hardware, modulation of power levels, and the placement of
] devices outside the trained workspace. Our results have
] interesting ramifications for location privacy in wireless
] networks.

We've actually done this at my work (AirDefense Inc.) and expect it to be in QA probably by Q1 of next year...its not that hard to train it if the system is clever about how it uses the data...

--Abaddon